Optimizing query processing using selectivity-awareness in Wireless Sensor Networks

Monitoring queries are fundamental for Wireless Sensor Networks (WSNs) that collect data for physical phenomena. In this work we address three key characteristics of monitoring queries. First, a monitoring query can be selective, i.e., it requests readings only from parts of a WSN. Second, a monitoring query can be continuous, i.e., it draws sensor readings for long periods of time. Finally, since physical phenomena are spatially correlated, a monitoring query selects spatially co-located nodes. In our earlier work, we proposed the Pocket Driven Trajectories (PDT) algorithm; a selectivity-aware data collection technique that tailors data collection paths for a monitoring query based on the spatial layout of selected nodes. In this work, we extend the basic PDT algorithm with an adaptive behavior. We show that the enhanced PDT algorithm is ideal for real world WSNs due to its two major strengths; the PDT algorithm is local, i.e., it does not require any global information about node locations or network connectivity. Furthermore, the PDT algorithm efficiently adapts its data collection paths over the lifetime of a query as changes in the spatial layout of selected nodes occur. Using extensive simulations, we show that in terms of energy efficiency the PDT algorithm clearly outperforms well-known WSN data collection algorithms.